tmap (version 1.2)

sample_dots: Sample dots from spatial polygons

Description

Sample dots from spatial polygons according to a spatial distribution of a population. The population may consist of classes. The output, a SpatialPointsDataFrame, can be used to create a dot map (see tm_dots), where the dots are colored according to the classes.

Usage

sample_dots(shp, vars = NULL, convert2density = FALSE, nrow = NA,
  ncol = NA, N = 250000, npop = NA, n = 10000, w = NA,
  shp.id = NULL, var.name = "class", var.labels = vars, unit = "km",
  unit.size = 1000, randomize = TRUE, ...)

Arguments

shp

A shape object, more specifically, a SpatialPolygonsDataFrame.

vars

Names of one or more variables that are contained in shp. If vars is not provided, the dots are sampled uniformly. If vars consists of one variable name, the dots are sampled according to the distribution of the corr

convert2density

Should the variables be converted to density values? Density values are used for the sampling algorithm, so use TRUE when the values are absolute counts.

nrow

Number of grid rows

ncol

Number of grid colums

N

Number of grid points

npop

Population total. If NA, it is recontructed from the data. If density values are specified, the population total is approximated using the polygon areas (see also unit and unit.size).

n

Number of sampled dots

w

Number of population units per dot. It is the population total divided by n. If specified, n is calculated accordingly.

shp.id

Name of the variable of shp that contains the polygon identifying numbers or names.

var.name

Name of the variable that will be created to store the classes. The classes are defined by vars, and the labels can be configured with var.labels.

var.labels

Labels of the classes (see var.name).

unit

Unit, see calc_densities. Needed to relate npop to w, if they are not both specified.

unit.size

Unit size, see calc_densities. Needed to relate npop to w, if they are not both specified.

randomize

should the order of sampled dots be randomized? The dots are sampled class-wise (specified by vars). If this order is not randomized (so if randomize=FALSE), then the dots from the last class will be drawn on top, which may intro

...

other arguments passed on to calc_densities and approx_areas

Details

The sampling algoritm is the following: TO DO

Examples

data(World)
World_dots <- sample_dots(World, vars="pop_est_dens", nrow=200, ncol=400, w=1e6)

tm_shape(World_dots) + tm_dots(size = .02, jitter=.1) + 
    tm_layout("One dot represents one million people", title.position = c("right", "bottom"))

# download Dutch neighborhood shape file
	dir <- tempdir()
	temp <- tempfile()
	
	download.file("http://www.cbs.nl/nl-NL/menu/themas/dossiers/nederland-regionaal/links/
				  2014-buurtkaart-shape-versie-1-el.htm", temp, mode="wb")
	unzip(temp, exdir = dir)
	
	NLD_nbhd <- read_shape(file.path(dir, "buurt_2014.shp"))
	
	# fix self-intersection but in shape object
	NLD_nbhd <- rgeos::gBuffer(NLD_nbhd, byid=TRUE, width=0)
	
	# remove all water neighborhoods
	NLD_nbhd <- NLD_nbhd[NLD_nbhd$OPP_LAND>0, ]
	
	# process data
	NLD_nbhd@data <- within(NLD_nbhd@data, {
		# convert land area ("OPP_LAND") from hectare (ha) to km2
		OPP_LAND <- OPP_LAND / 100
		
		# set negative percentages of western and non-western immigrants to 0
		P_WEST_AL[P_WEST_AL<0] <- 0
		P_N_W_AL[P_N_W_AL<0] <- 0
		
		# calculate population density values
		dens <- (AANT_INW / OPP_LAND)
		dens[is.nan(dens)] <- 0
		
		# divide population in three groups: 
		# western and non-western immigrants, and native Dutch
		west <- dens * P_WEST_AL / 100
		non_west <- dens * P_N_W_AL / 100
		dutch <- dens - non_west - west
	})
	
	# Select The Hague (Den Haag) area
	DH_bbox <- bb(NLD_nbhd[which(NLD_nbhd$GM_NAAM=="'s-Gravenhage"), ])
	
	# Crop shape to The Hague area (use 1.05 bounding box extension to make sure the
	# whole region is covered after reprojection to the Mercator porjection)
	DH_nbhd <- raster::crop(NLD_nbhd, bb(DH_bbox, 1.05))
	
	# Read OSM layer
	DH_nbhd_osm <- read_osm(bb(DH_bbox, current.projection="rd", projection="longlat"), 
							type = "mapquest")
	
	# Sample dots (each dot represents 100 persons)
	DH_nbhd_dots <- sample_dots(DH_nbhd, c("dutch", "west", "non_west"), 
		convert2density = FALSE, 
		N=250000, w=100, 
		var.labels = c("Dutch (native)", "Western immigrants", "Non-western immigrants"), 
		shp.id = "ID")
	
	# Show map
	tm_shape(DH_nbhd_osm) + tm_raster(saturation=.2) +
		tm_shape(DH_nbhd_dots) + 
		    tm_dots("class", size=.04, alpha=.75, palette="Dark2", 
		    		title = "The Hague population") +
		tm_layout(inner.margins=0, legend.frame=TRUE, legend.bg.color="grey90")